Keynote: Linus Torvalds, Creator of Linux & Git, in Conversation with Dirk Hohndel
Linus Torvalds discusses AI’s effect on open-source infrastructure, including crawler strain on kernel.org and AI-generated spam reports. He is skeptical of ‘vibe coding’ for serious maintenance but sees AI as a learning gateway for beginners. He compares AI to compilers, a tool that removes minutiae rather than replacing developers, and predicts more software jobs, not fewer.
AI can make a first working program easier to create without making a long-lived project easy to maintain. In conversation with Dirk Hohndel, Linus Torvalds welcomes that lower entry barrier while describing the costs AI crawlers and fabricated reports already impose on open-source maintainers.
AI's Impact on Open Source Infrastructure
- Crawlers consume infrastructure capacity: Torvalds says AI crawlers scanning
kernel.orghave disrupted its source infrastructure 15:25–16:28(opens in a new tab). - Fabricated reports consume maintainer time: He describes made-up bug and security reports from people misusing AI. Hohndel raises curl's experience; Torvalds says the kernel has seen the same problem, but less severely 16:28–17:33(opens in a new tab).
"Vibe Coding" & AI as a Tool
- A demo is not a maintained product: Torvalds says he does not use coding agents himself. He suspects prompt-led “vibe coding” may be a poor fit for long-term maintenance. Hohndel says agents get him 90% of the way; Torvalds replies that the last 10% can take 34 years of a 35-year project 18:23–21:31(opens in a new tab).
- Small projects can draw beginners into programming: Torvalds compares agent-built personal programs with typing programs from magazines when he was young. Lowering the initial hurdle can help people get excited about computers, even if the result is not ready to become a durable product 18:23–20:27(opens in a new tab).
- Kernel assistance remains experimental: He describes work on helping maintainers process patches and backport them to stable versions, rather than claiming established use for complex kernel code generation 15:25–16:28(opens in a new tab).
The Future of Software Jobs
Torvalds says he does not know what will happen to software jobs. His “gut feel” is that AI will resemble compilers: it removes some low-level work while leaving a need for programmers and maintainers. He thinks greater productivity might also open enough new areas of development to create more programming work 21:31–23:39(opens in a new tab).
Hardware & Industry Shifts
- AI gives Nvidia a stronger reason to support Linux: Torvalds says Linux's importance to AI infrastructure has made Nvidia a more constructive kernel contributor 14:22–15:25(opens in a new tab).
- The kernel still supports a general-purpose system: He remains most interested in the CPU and sees proprietary AI workloads as another kind of application running above Linux, much like commercial databases 11:26–14:22(opens in a new tab).
Full video: https://www.youtube.com/watch?v=tWx769t1JKg(opens in a new tab)